This is an archive of the discontinued LLVM Phabricator instance.

clang/lib/Driver/ToolChains/Clang.cpp
7215	will do. Since this option is shared between all AMDGPU toolchains, I will add it to AMDGPUToolChain::addClangWarningOptions and let the derived toolchains call it.
clang/test/Driver/hip-options.hip
144	will fix

revised by Artem's and Fangrui's comments

Herald added a reviewer: jdoerfert. · View Herald TranscriptJun 7 2023, 7:48 AM

Herald added subscribers: jplehr, sstefan1. · View Herald Transcript

Harbormaster completed remote builds in B237270: Diff 529301.Jun 7 2023, 8:41 AM

Revision Contents

Path

Size

clang/

lib/

Driver/

ToolChains/

Clang.cpp

4 lines

test/

Driver/

hip-options.hip

7 lines

Diff 528937

clang/lib/Driver/ToolChains/Clang.cpp

This file is larger than 256 KB, so syntax highlighting is disabled by default.

	Show First 20 Lines • Show All 7,203 Lines • ▼ Show 20 Lines

	if (Triple.isAMDGPU()) {			if (Triple.isAMDGPU()) {
	handleAMDGPUCodeObjectVersionOptions(D, Args, CmdArgs);			handleAMDGPUCodeObjectVersionOptions(D, Args, CmdArgs);

	Args.addOptInFlag(CmdArgs, options::OPT_munsafe_fp_atomics,			Args.addOptInFlag(CmdArgs, options::OPT_munsafe_fp_atomics,
	options::OPT_mno_unsafe_fp_atomics);			options::OPT_mno_unsafe_fp_atomics);
	Args.addOptOutFlag(CmdArgs, options::OPT_mamdgpu_ieee,			Args.addOptOutFlag(CmdArgs, options::OPT_mamdgpu_ieee,
	options::OPT_mno_amdgpu_ieee);			options::OPT_mno_amdgpu_ieee);

				// AMDGPU does not support atomic lib call. Treat atomic alignment
				// warnings as errors.
				CmdArgs.push_back("-Werror=atomic-alignment");
				traUnsubmitted Done Reply Inline Actions Should it be done from `HIPAMDToolChain::addClangWarningOptions` ? That's where Darwin does similar propotion from a waring to an error. tra: Should it be done from `HIPAMDToolChain::addClangWarningOptions` ? That's where Darwin does…
				yaxunlAuthorUnsubmitted Done Reply Inline Actions will do. Since this option is shared between all AMDGPU toolchains, I will add it to AMDGPUToolChain::addClangWarningOptions and let the derived toolchains call it. yaxunl: will do. Since this option is shared between all AMDGPU toolchains, I will add it to…
	}			}

	// For all the host OpenMP offloading compile jobs we need to pass the targets			// For all the host OpenMP offloading compile jobs we need to pass the targets
	// information using -fopenmp-targets= option.			// information using -fopenmp-targets= option.
	if (JA.isHostOffloading(Action::OFK_OpenMP)) {			if (JA.isHostOffloading(Action::OFK_OpenMP)) {
	SmallString<128> Targets("-fopenmp-targets=");			SmallString<128> Targets("-fopenmp-targets=");

	SmallVector<std::string, 4> Triples;			SmallVector<std::string, 4> Triples;
	▲ Show 20 Lines • Show All 1,362 Lines • Show Last 20 Lines

clang/test/Driver/hip-options.hip

	Show First 20 Lines • Show All 132 Lines • ▼ Show 20 Lines

	// RUN: %clang -### -nogpuinc -nogpulib -mamdgpu-ieee -mno-amdgpu-ieee -ffast-math \			// RUN: %clang -### -nogpuinc -nogpulib -mamdgpu-ieee -mno-amdgpu-ieee -ffast-math \
	// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 \| FileCheck -check-prefixes=IEEE-OFF %s			// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 \| FileCheck -check-prefixes=IEEE-OFF %s
	// IEEE-OFF: clang{{.}} "-triple" "amdgcn-amd-amdhsa" {{.}} "-mno-amdgpu-ieee"			// IEEE-OFF: clang{{.}} "-triple" "amdgcn-amd-amdhsa" {{.}} "-mno-amdgpu-ieee"

	// RUN: %clang -### -nogpuinc -nogpulib -mamdgpu-ieee -mno-amdgpu-ieee -ffast-math \			// RUN: %clang -### -nogpuinc -nogpulib -mamdgpu-ieee -mno-amdgpu-ieee -ffast-math \
	// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 \| FileCheck -check-prefixes=IEEE-OFF-NEG %s			// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 \| FileCheck -check-prefixes=IEEE-OFF-NEG %s
	// IEEE-OFF-NEG-NOT: clang{{.}} "-triple" "amdgcn-amd-amdhsa" {{.}} "-mamdgpu-ieee"			// IEEE-OFF-NEG-NOT: clang{{.}} "-triple" "amdgcn-amd-amdhsa" {{.}} "-mamdgpu-ieee"

				// Check -Werror=atomic-alignment is passed for amdpu by default.

				// RUN: %clang -### -target x86_64-unknown-linux-gnu -nogpuinc -nogpulib \
				MaskRayUnsubmitted Done Reply Inline Actions Prefer `--target=` to deprecated (since 3.x) `-target` MaskRay: Prefer `--target=` to deprecated (since 3.x) `-target `
				yaxunlAuthorUnsubmitted Done Reply Inline Actions will fix yaxunl: will fix
				// RUN: --cuda-gpu-arch=gfx906 %s 2>&1 \| FileCheck -check-prefix=WARN-ATOMIC %s
				// WARN-ATOMIC: clang{{.}} "-triple" "amdgcn-amd-amdhsa" {{.}} "-Werror=atomic-alignment"
				// WARN-ATOMIC-NOT: clang{{.}} "-triple" "x86_64-unknown-linux-gnu" {{.}} "-Werror=atomic-alignment"

This is an archive of the discontinued LLVM Phabricator instance.

[AMDGPU] Diagnose unaligned atomic for amdgpuNeeds ReviewPublic

Details

Diff Detail

Event Timeline

Revision Contents

Diff 528937

clang/lib/Driver/ToolChains/Clang.cpp

clang/test/Driver/hip-options.hip

[AMDGPU] Diagnose unaligned atomic for amdgpu
Needs ReviewPublic